Introduction

Over 790 000 people are diagnosed with acute myocardial infarction (AMI, often referred to as heart attack) annually in America alone1. Diagnosis as early and accurate as possible warrants immediate medical care, which would reduce mortality and improve the prognosis of AMI2,3. AMI is classified into ST-segment elevation myocardial infarction (STEMI) and non-STEMI (NSTEMI)4. For STEMI, reperfusion therapy should be administered as quickly as possible to reduce infarct size and mortality5. For NSTEMI and unstable angina (UA), another non-ST-segment elevation acute coronary syndrome, detailed revascularization strategies were recommended based on the clinical features of individual patients3. Thus, STEMI patients require early diagnosis, whereas NSTEMI and UA patients require detailed clinical features. Hence, it is critical to build a nexus of molecular biomarkers for physicians to organize an effective, personalized therapeutic schedule.

Biomarker tests combined with electrocardiographic (ECG) analysis is the main tactic for AMI diagnosis. Although ECG provide strong evidence for ischemia or infarction, ECG alone is incapable of diagnosing NSTEMI, which comprises 60%–75% of all myocardial infarctions6. In addition, both STEMI and NSTEMI demand biomarkers for final diagnosis according to European and American guidelines4,6. Biomarkers for AMI should first be quantitatively altered in AMI and thus be used to predict and monitor the pathogenic processes of AMI7,8. Meanwhile, a good biomarker should also be stable and easily accessible9. The currently preferred diagnostic biomarkers for AMI are cardiac troponin I and T (cTnI and cTnT)10. Circulating cTnI and cTnT have been the 'gold standard' of AMI diagnosis for over 20 years since they can be released from necrotic cardiomyocytes within 2–4 h post AMI10,11. Circulating cTnI and cTnT reach peak levels at 24-48 h post AMI and last for over a week12. High-sensitive cardiac troponin I and T were recently developed to improve the sensitivity and accuracy of AMI diagnosis13,14. However, false positive results with elevated cTn occurs in patients with heart failure, chronic kidney diseases and sepsis, especially in elderly patients15,16,17,18. In addition, since cTn remains in the circulation for over 7 d12, small repeat infarctions post major infarction are unlikely detectable. Thus, it is critical to identify sensitive biomarkers for extremely early diagnosis of STEMI as well as specific biomarkers that monitor the entire pathogenic processes of AMI.

Starting with the exploration of microRNAs, numerous efforts have been made to excavate the treasures underlying non-coding RNAs (ncRNAs), which consist of over ninety percent of the human genome. MicroRNAs (miRNAs), small interference RNAs (siRNAs), long non-coding RNAs (lncRNAs) and recently identified circular RNAs (circRNAs) have all been shown to have regulatory functions or diagnostic potential in cardiovascular diseases19. MicroRNAs are endogenous short non-coding RNAs with a length of approximately 22 nt expressed in almost every cell20. Mature miRNAs bind to RNA binding proteins (RBPs) and regulate mRNA stability via recognition sites, such as AU-rich elements, in the untranslated regions of mRNA21,22,23. In 2008, scientists in the UK24, USA25 and China26 independently found that circulating microRNAs are sensitive biomarkers for cancer and other kinds of disease. Pioneering works have demonstrated that microRNAs possess huge potential as biomarkers for AMI27,28. Recently, other classes of non-coding RNAs, such as lncRNAs and circRNAs, have also been suggested as biomarkers of AMI. Despite accumulating reports on non-coding RNAs as biomarkers of AMI, we summarize recent findings and discuss future applications of non-coding RNAs as biomarkers of acute myocardial infarction with special attention to the diagnostic value and prognostic potential of these non-coding RNAs.

Non-coding RNAs as diagnostic biomarkers for AMI

MicroRNAs

The first group of miRNAs identified as biomarkers for AMI patients were miR-208a, miR-499, miR-13327 and miR-127,28. In the plasma of STEMI patients, miR-208a, miR-1, miR-499 and miR-133 were significantly increased. Among these miRNAs, miR-208a was expressed specifically in cardiomyocytes and more importantly, miR-208a showed 90.9% sensitivity and 100% specificity for AMI diagnosis, representing a more advantageous biomarker than even cTnI27. As we mentioned earlier, it is critical to identify new biomarkers for extremely early STEMI diagnosis to remedy the delayed plasma peak of cTn. Importantly, both rat models and mice models were utilized to screen for extremely early diagnostic miRNAs27,29. In rat AMI models, miR-208a increased by more than 50 fold within 1 h post coronary artery occlusion surgery and reached a peak level at 3 h after surgery, displaying kinetics that peaked just before a significant elevation in cTn27. In addition, miR-1 showed an over 200-fold increase in rat AMI models at 6 h post surgery30. A most recent study that outlines the release kinetics of circulating miR-208a in AMI patients further supported the suggestion that miR-208a is a promising diagnostic index for early diagnoses of AMI31. Even though the increase in miR-499 was not as dramatic as that of miR-208a, plasma miR-499 levels were significantly increased in rat AMI models 1 h post AMI27. A subsequent report further demonstrated that miR-208b and miR-499 were released into the circulation upon various types of myocardial damage, including AMI, and miR-208b showed an over 1600-fold increase in AMI patients32. In addition, a cohort with one thousand patients validated the increase in circulating miR-208b and miR-499 in patients with chest pain33.

The functions of miR-208a, miR-1, miR-499 and miR-133 have been well characterized in the heart34,35,36,37,38. miR-208a was initially identified in heart tissues and regulates the proper expression of β-MHC36,39. In addition, miR-208 regulates cardiomyocyte apoptosis via Bcl-2 and Bax signaling40,41. Most recently, miR-208a-3p was found to aggravate autophagy through the PDCD4-ATG5 pathway42. Recently, the diagnostic values of miR-208a, miR-1, miR-499 and miR-133 in the circulation have been intensively studied. In addition to mouse and rat AMI models, pig AMI models were established, and miR-208a, miR-1, miR-499-5p and miR-133a were all found to be elevated in pig plasma within 2 h post AMI43. Interestingly, in a pig AMI model and a rat AMI model, miR-1 was significantly increased in urine, and further analysis showed that miR-1 was also detectable in the urine of patients with AMI43,44. Recently, a study in a Chinese Han population with plasma samples drawn from patients less than 2 h after the onset of symptoms further suggested the potential of rapid diagnosis with these miRNAs45. All these reports validated that circulating miR-208a, miR-1, miR-499 and miR-133 are bona fide biomarkers for AMI. Moreover, researchers found that these cardiomyocyte-enriched miRNAs are capable of differentiating STEMI from NSTEMI. In a 444-patient cohort, miR-133a was found to distinguish STEMI, NSTEMI and UA, and miR-208a was able to distinguish STEMI from NSTEMI, whereas miR-499 and miR-208b were not able to distinguish between STEMI and NSTEMI46. However, in another cohort, miR-499 and miR-208b were found to be able to distinguish STEMI from NSTEMI47. In addition, miR-499-5p was specifically highly sensitive as a NSTEMI biomarker in elderly patients48. Additionally, miR-133a displayed distinguishable circulating levels in the serum samples of STEMI and NSTEMI patients, and serum miR-133a also showed diagnostic value for UA and takotsubo cardiomyopathy (TTC)49. More importantly, since neither ECG nor cTn is able to differentiate TTC from STEMI, it is exciting that both serum and plasma miR-133a were able to differentiate TTC from STEMI49,50. In addition, miR-499 was identified as a signature for perioperative myocardial infarction in patients undergoing coronary artery bypass graft surgery51,52. All these data indicated that the first group miRNAs identified as biomarkers for AMI possess huge potential in clinical applications, and precise interpretations of these results are promising for Precision Medicine (Figure 1).

Figure 1
figure 1

Building a miRNA release spectrum after AMI. With current and future information on circulating miRNAs after the onset of AMI, we can build a miRNA release spectrum that contains detailed messages on when and how much miRNA is released into the circulation after AMI. With this spectrum, physicians can make extremely early diagnosis of STEMI or understand the status of an AMI patient that is either suffering from a STEMI or enduring a NSTEMI.

Full size image

The number of novel miRNAs used as diagnostic biomarkers for AMI has increased rapidly. Although the sensitivity and accuracy of these circulating miRNAs vary, several novel circulating miRNAs with special release kinetics are promising candidates. Two circulating miRNAs, miR-30a and miR-195, with special kinetics post AMI were observed. These two miRNAs increased significantly at only 8 h post AMI and rapidly dropped back to baseline53. Circulating miR-122-5p also showed similar kinetics to miR-30a and miR-195, which peaked at 8 h post AMI54,55. However, circulating miR-122-5p did not show a quick drop at 12 h post AMI, but miR-30a and miR-195 did. Another circulating miRNA with unique kinetics post AMI was circulating miR-181a; its level in the circulation increased from 6 h post AMI and reached peak at 24 h post AMI56. These results suggested an option that various miRNAs with diverse releasing kinetics could be integrated together and work as monitors of the pathogenic processes of AMI. miR-328 was found increased by over 10 fold in the plasma of AMI patients in a small sample cohort57, and a large sample cohort (carried out later) further validated the over 10-fold increase of miR-328a in the plasma of AMI patients and proposed that miR-134 is another circulating miRNA that has potential in AMI diagnosis58. Although another report also suggested a rise in circulating miR-134, both studies showed that only an approximately 5-fold increase was observed post AMI58,59. Atherosclerosis-related miR-21-5p and miR-361-5-p in the circulation were increased in AMI patients as well as in patients with ischemic stroke60. Novel circulating miRNA were also identified to differentiate STEMI from NSTEMI. Circulating miR-1291 showed a significant increase in NSTEMI compared to STEMI, whereas circulating miR-663b was 2-fold higher in STEMI than in NSTEMI61. In addition, circulating miR-486 and miR-150 also varied from STEMI to NSTEMI62. Also, circulating miR-125b-5p and miR-30d-5p distinguished AMI from UA63. These data suggested that a combination of these circulating miRNAs could provide a biomarker signature to differentiate STEMI from NSTEMI and UA. Circulating miR-19a and miR-19b also showed modest increases in the plasma of AMI patients59,64. Interestingly, the circulating level of miR-22, a miRNA participating in cardiac hypertrophy65, is also elevated in the blood of AMI patients66.

Multiple novel circulating miRNAs that decreased in AMI patients have also been identified. Circulating let-7b and miR-126 were found to be dramatically decreased to 1/10 in the plasma of AMI patients from as early as 4 h post onset of AMI53,67. Moreover, the serum level of miR-126 was also decreased in AMI patients68. Since both let-7b and miR-126 play critical roles in heart development and heart diseases, it is of great importance to further investigate the biological significance of the decrease of these two circulating miRNAs69,70,71,72,73. Circulating miR-519e-5p and miR-99a were also decreased to approximately 50% in AMI patients60,74. In addition, plasma miR-145 was found to be reduced in both NSTEMI and STEMI patients and patients with heart failure75. With the fast development of high-throughput miRNA detection technologies, additional novel decreased circulating miRNAs were easily identified in AMI patients76,77,78. Circulating miR-320b and miR-125b were found to be decreased with microarray assay in the plasma of AMI patients, and these reductions were further validated in a cohort of 178 Chinese AMI patients77. Another microarray-based study showed that plasma miR-26a and miR-191 were decreased in AMI patients, which is in line with a previous report suggesting that serum miR-26a and serum miR-191 were decreased in AMI patients68,78. Notably, miR-155 and miR-126, two miRNAs critical for vascular biology69,70,79, were both decreased in the plasma of patients post off-pump coronary artery bypass graft, suggesting a potential monitoring function of miR-155 and miR-126 in cardiac surgery51. Taken together, these data suggested that by building up a thorough network of circulating miRNAs post AMI, it is possible to diagnose STEMI rapidly and monitor the pathogenic processes of AMI and thus to organize personalized therapeutic schedules for patients with a risk of AMI.

Long non-coding RNAs and circular RNAs

Recently, non-coding RNA other than miRNAs, such as long non-coding RNAs and circular RNAs, have drawn tremendous attention from scientists in the cardiovascular biology field80,81,82. Circulating forms of these non-coding RNA are also potential biomarkers for cardiovascular diseases19,83. Here, we summarize recent research on non-coding RNAs (other than miRNAs) as diagnostic biomarkers of AMI.

The first study on circulating lncRNAs tested the expression levels of five lncRNAs in the blood cells of 414 patients and 86 healthy controls84. Four out of the five lncRNAs, hypoxia inducible factor 1A antisense RNA 2 (aHIF), cyclin-dependent kinase inhibitor 2B antisense RNA 1 (ANRIL), potassium voltage-gated channel, KQT-like subfamily, member 1 opposite strand/antisense transcript 1 (KCNQ1OT1) and metastasis-associated lung adenocarcinoma transcript 1 (MALAT1), are differentially expressed in AMI patients and more importantly, MALAT1, ANRIL, KCNQ1OT1 and MI associated transcript (MIAT) were able to distinguish STEMI from NSTEMI84. Two more lncRNAs, Zinc finger antisense 1 (ZFAS1) and Cdr1 antisense (CDR1AS), were identified with altered expression levels in the whole blood of AMI patients85. Another study showed alterations of over 5000 lncRNAs in the whole blood of Chinese Uyghur AMI patients86. In addition, lncRNA Urothelial carcinoma-associated 1 (UCA1), which is detectable in plasma, decreased in AMI patients at two hours after the onset of symptoms87. Although these studies provided excellent scientific significance, the sensitivity of these lncRNAs as biomarkers was not comparable to the miRNAs identified as diagnostic biomarkers for AMI. The expression of ZFAS1, CDRAS and MIAT were also affected in a MI mice model, and attenuation of MIAT abrogated the fibrogenesis induced by MI85,88. CircRNAs are a covalently closed circular form of non-coding RNA that presents in various organisms89,90. Due to their specific structure, circRNAs have great potential as biomarkers in disease diagnosis91. CircRNA Cdr1as was elevated in myocardial tissues from MI mice, suggesting that circRNAs may also have potential as diagnostic biomarkers for AMI92. In addition, a circRNA microarray was applied to detect altered circRNAs in MI mice, which further provided bases for the application of circRNAs as biomarkers93.

In summary, several attempts to use non-coding RNAs, especially miRNAs, as diagnostic biomarkers for AMI have become milestones in biomarker identification. Researchers and clinicians are paying more attention to uniting the protocols for sample collection and management. Thus, specific novel non-coding RNAs with specific diagnostic values can be expected in the future. Nonetheless, a handful of circulating miRNAs have already shown better qualities than traditional biomarkers, such as cTn. Further kinetic characterization of these miRNAs will provide detailed information for AMI diagnosis.

Non-coding RNAs as prognostic biomarkers post AMI

Non-coding RNAs as prognostic biomarkers for cardiac death post AMI

Since miRNAs are relatively stable due to various regulatory mechanisms94,95, researchers are not satisfied with using circulating miRNAs as only diagnostic biomarkers for AMI. Hence, ever since their identification, efforts have been put into using circulating miRNAs as prognostic biomarkers post AMI.

Among the first group of circulating miRNAs identified in AMI patients, miR-133a and miR-208b were first associated with mortality46. High levels of circulating miR-133a and miR-208b were found to be associated with all-cause mortality at 6 months in a 444-patient cohort46. The association between increased circulating miR-208b levels and mortality was subsequently validated by other AMI cohorts96,97. Moreover, high plasma levels of miR-499 were also associated with 30-d, 4-month, 1-year, 2-year and 6-year mortality96,97,98. Based on a miRNA array, increased levels of serum miR-155 and miR-380 were strongly associated with cardiac death99. More importantly, miR-192, miR-194 and miR-34 were significantly higher in the serum collected from a group of patients who developed ischemic heart failure after AMI onset100. In addition, high levels of serum miR-145 were also associated with cardiac death and heart failure in post AMI patients101. Recently, it was proposed that the ratio of serum miR-122-5p/133b measured at the time of cardiac catheterization was a strong predictor of cardiac death102. Although future investigations for the validation of these miRNAs as prognostic biomarkers post AMI are necessary, it is undeniable that aberrant levels of circulating miRNAs provided predictive information on adverse cardiovascular events post AMI, such as cardiac death and heart failure.

Non-coding RNAs as prognostic biomarkers for cardiac function post AMI

Despite improvements in reperfusion rates and secondary preventive medications post AMI, cardiac dysfunction, especially left ventricular dysfunction, developed at a relatively high rates103,104. Thus, it is of special importance to identify biomarkers that predict cardiac dysfunction in patients who suffered from AMI and subsequently provide medical suggestions for these patients. Based on the valuable information behind circulating non-coding RNAs in both diagnostic and prognostic levels for AMI patients, increasing studies have focused on the prognostic value of circulating non-coding RNAs for cardiac function post onset of AMI.

In addition to the roles in AMI diagnosis and cardiac death prediction post AMI, high serum levels of miR-133a were found to be associated with decreased myocardial salvage, larger infarcts and graver reperfusion injury in AMI patients post reperfusion105. In addition, elevated plasma levels of miR-133a were associated with coronary artery stenosis in patients with coronary heart disease106. Circulating miR-1, miR-208b and miR-499 were found to be negatively associated with left ventricular ejection fraction (LVEF) in patients treated with percutaneous coronary intervention96. In addition to miRNAs, high levels of circulating lncRNA MALAT1 also predicted poor LVEF in AMI patients four months post reperfusion84. Circulating lncRNA were also proposed to have roles in cardiac fibrosis post infarction88,107. Left ventricular (LV) remodeling is considered to be a predominate cause of heart failure in AMI patients108. By large-scale screening, several plasma miRNAs, especially decreased plasma miR-150, were found to be strongly associated with LV remodeling109. It was further characterized that a panel of 4 circulating miRNAs, miR-16/miR-27a/miR-101/miR-150, predicted LV remodeling after AMI with a net reclassification improvement of 66%110. In addition, increased circulating levels of miR-208b, miR-34a, miR-21 and miR-155 were all associated with LV remodeling after AMI111,112,113. Importantly, a circular RNA, MICRA, was identified recently as a circulating biomarker that predicts LV dysfunction, which expanded our understanding of circular RNAs114.

Taken together, circulating non-coding RNAs showed promising roles as prognostic biomarkers post AMI. Future validations of the previous findings in larger cohorts and standardizations of detection schemes could further promote the recognition and application of non-coding RNAs as biomarkers for risk stratification of AMI patients.

Non-coding RNAs as risk predictive biomarkers for AMI

Two studies with unique significance based on large biobanks identified circulating miRNAs that predict future myocardial infarction in healthy individuals115,116. Nineteen circulating miRNAs of 820 participants of the Bruneck cohort were quantified, and the association between levels of these circulating miRNAs and myocardial infarction incidence were estimated. In the end, researchers found that increased levels of circulating miR-126 and decreased levels of circulating miR-223 and miR-197 were associated with incidence of myocardial infarction in apparently health participants115. Based on the HUNT (The Nord-Trøndelag Health Study) report, a combination of 5 miRNAs, miR-106-5p/miR-424-5p/let-7g-5p/miR-144-3p/miR-660-5p, correctly predicted 77.6% incidence of myocardial infarction in apparently health participants. More importantly, by adding this miRNA combination to the Framingham Risk Score, the AUC increased significantly from 0.72 to 0.91116. These two illuminating studies indicated a potential use of miRNAs as biomarkers to predict heart attack for apparently healthy people. With the popularization of physical examinations and the development of personalized medicine, the establishment of miRNAs as biomarkers to predict the incidence of heart attacks is evolutionary meaningful.

Conclusion and perspectives

Former genomic 'garbage' has now proven itself as a powerful regulator in every biological process. The identification of each type of non-coding RNAs has led a revolution not only in our understanding of the molecular network but also in clinical applications. Multiple miRNAs have exhibited excellent properties as biomarkers for cancers, neurological diseases and cardiovascular diseases117,118,119,120,121. Cardiac-specific miR-208a is no doubt the most promising STEMI biomarker thus far. Not only because the specificity and sensitivity of miR-208a are outstanding but also because the kinetics of miR-208a release showed superiority over the 'gold standard' cTn: 1) miR-208a can be detected within 2 h, which is earlier than cTn, and allows extremely early diagnosis of AMI; and 2) miR-208a declined to baseline within 24 h, which enables the detection of minor cardiac events post major infarction. In addition, based on existing reports, we are able to build a miRNA release spectrum (Figure 1) after AMI, which provides quantitative and time-lapse monitoring of AMI progress. With this spectrum, we are able to customize personal medical care, which prevents further damage.

Despite what researchers have accomplished, validations of non-coding RNAs with dramatic changes and swift responses to the onset of AMI in larger cohorts are imperatively needed. Meanwhile, the establishment of standardized detection methods and release kinetics of novel non-coding RNAs are equally important. To finally facilitate the clinical applications of non-coding RNAs as biomarkers for AMI, future studies should focus on the following: 1) patient information, especially the detailed time of symptom onset, which helps identify extremely early biomarkers for STEMI; 2) inclusion of multiple examination time points, which enables long-term monitoring of the course of NSTEMI or UA; and 3) collaboration to increase cohort size and unify detection methods, which improves the applicability of certain biomarkers to a wider population. Meanwhile, it is of significance to further analyze the predictive value of non-coding RNAs for incident myocardial infarction of apparently healthy people.

Even though several animal models have been utilized to describe the accurate release kinetics of these non-coding RNAs that may serve as biomarkers for AMI, elaborate release kinetics that reflect the authentic release kinetics of AMI patients are still not available. A nonhuman primate AMI model may provide infusive results because 1) nonhuman primates are so close to human beings that they share highly conserved genetic regulation networks; thus, all kinds of non-coding RNA, even lncRNA, could be exploited. 2) As a big animal model, primates have enough blood for time-lapse sampling. Thus, it is of great potential to utilize a primate AMI model to further explore the applications of non-coding RNAs as biomarkers for AMI.

In summary, circulating non-coding RNAs, especially miRNAs, displayed numerous advantages as biomarkers and built a multi-level nexus that illustrates that the biological nature of AMI is a promising prospect for AMI diagnosis and prognosis.